1964 Binford L. A Consideration of Archaeological Research Design

A Consideration of Archaeological Research Design
Author(s): Lewis R. Binford

Source: American Antiquity, Vol. 29, No. 4 (Apr., 1964), pp. 425-441
Published by: Cambridge University Press
Stable URL: http://www.jstor.org/stable/277978
Accessed: 02-03-2018 12:44 UTC
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms
Cambridge University Press is collaborating with JSTOR to digitize, preserve and extend
access to American Antiquity
This content downloaded from 200.21.104.18 on Fri, 02 Mar 2018 12:44:47 UTC
All use subject to http://about.jstor.org/terms
AMERICAN ANTIQUITY
VOLUME 29 APRIL 1964 NUMBER 4
A CONSIDERATION OF ARCHAEOLOGICAL RESEARCH DESIGN*

LEWIS R. BINFORD
ABSTRACT 4). A cultural system is a set of constant or

It is argued that the methodology most appropriate cyclically repetitive articulations between the so-
for the task of isolating and studying processes of cul-
cial, technological, and ideological extrasomatic,
tural change and evolution is one which is regional inadaptive means available to a human population
scope and executed with the aid of research designs
based on the principles of probability sampling. The (White 1959: 8). The intimate systemic articu-
lation of localities, facilities, and tools with
various types of observational populations which archae-
ologists must study are discussed, together with an evalu-specific tasks performed by social segments re-
ation of the methodological differences attendant upon sults in a structured set of spatial-formal rela-
adequate and reliable investigation of each. Two basic
sampling universes are discussed, the region and the site,
tionships in the archaeological record. People
together with their methodological and research-design do not co-operate in exactly the same way when
performing different tasks. Similarly, different
peculiarities. These are used as a basis for discussion and
past and current research programs are evaluated in terms
tasks are not uniformly carried on at the same
of what are believed to be major limitations in obtaining
the "facts" pertinent to studies of cultural processes.
locations. As tasks and co-operating groups vary,
so do the implements and facilities (Wagner
IT SEEMS FAIR to generalize that archaeolo- 1960: 88-117) of task performance. The loss,
gists are becoming more interested in the ex-breakage, and abandonment of implements and
planatory potential which studies of paleoecol-facilities at different locations, where groups of
ogy, paleodemography, and evolution offer for variable structure performed different tasks,
increasing our understanding of formal and leaves a "fossil" record of the actual operation
structural change in cultural systems. Several of an extinct society. This fossil record may be
read in the quantitatively variable spatial clus-
anthropologists have recognized a growing inter-
terings of formal classes of artifacts. We may not
est in questions dealing with the isolation of con-
always be able to state or determine what spe-
ditions and mechanisms by which cultural
cific activities resulted in observed differential
changes are brought about (Adams 1960; Braid-
distributions, but we can recognize that activities
wood 1959; Haag 1959; Steward 1960). In short,
were differentiated and determine the formal na-
we seek answers to some "how and why" ques-
ture of the observable variability. I have argued
tions in addition to the "what, where, and when"
elsewhere (Binford 1962: 219) that we can re-
questions so characteristically asked by archae-
cover, both from the nature of the populations of
ologists. This paper is concerned with present-
artifacts and from their spatial associations, the
ing certain methodological suggestions, some of
fossilized structure of the total cultural system.
which must be adopted if we are to make prog-
The archaeological structure of a culture should,
ress in the study of processes and move archae-
and in my opinion does, reflect all other struc-
ology into the "explanatory level" of develop-
tures, for example, kinship, economic, and politi-
ment (Willey and Phillips 1958: 4-5).
cal. All are abstracted from the events which
In any general discussion of method and
occur as part of the normal functioning of a cul-
theory there is inevitably an argumentative bias
tural system. The archaeological structure results
on the part of the writer. It should be pointed
from these same events. The definition of this
out that I believe the isolation and study of cul-
tural systems, rather than aggregates of culture structure and the isolation of the archaeological
traits, is the only meaningful approach to under- remains of a cultural system are viewed as re-
standing cultural processes (Steward 1960: 173- search objectives. Such an isolation can be made
by the demonstration of consistent between-class
* This paper was presented at the Annual Meeting of correlations and mutual co-variations among
the Society for American Archaeology at the University
of Colorado, Boulder, Colorado, May, 1963. classes of artifacts and other phenomena. The
425
426 AMERICAN ANTIQUITY [ VOL. 29, No. 4, 1964
isolation and definition of extinct cultural extent

sys- of such regions will vary because it is
tems, both in terms of content and demonstra-recognized that cultural systems differ greatly in
ble patterns of mutual formal-spatial co-varia-
the limits of their adaptive range and milieu.
tion, can be accomplished. Once accomplished,As cultural systems become more complex, they
such an archaeological structure is amenable to
generally span greater ecological ranges and
analysis in terms of form and complexity; in into more complex, widespread, extraso-
enter
short, we can speak of culture types. Methodscietal interaction. The isolation and definition
for correlating archaeologically defined culture
of the content, the structure, and the range of
types with structural forms defined in termsa of
cultural system, together with its ecological
behavioral attributes can be developed. When relationships, may be viewed as a research ob-
this is accomplished, archaeologists and "social
jective. Admittedly it is an objective which
anthropologists" will be in the position to make
may or may not be successfully accomplished
joint contributions to the solution of commonunder any given research design. The research
anthropological problems, a condition that hard-
design should be aimed at the accomplishment
ly obtains today. of this isolation which, I believe, is most profit-
In addition to maintaining the position that ably prosecuted within a regional unit of investi-
we should strive to isolate the archaeological gation. Under current programs of salvage ar-
chaeology and increased foundation support for
structure of extinct cultural systems, it is argued
archaeological research, we are being given the
that changes in cultural systems must be investi-
gated with regard to the adaptive or coping situ-
opportunity to study major regions intensively.
ations which are presented to human popula- In spite of the opportunities currently available,
tions. If we are profitably to study process,it we
is my impression that very little thought has
must be able to isolate cultural systems been and given to research design. Methods and
study them in their adaptive milieu conceived approaches utilized in such investigations seem
in terms of physical, biological, and socialtodi- be little more than expanded or greatly en-
mensions. The physical and biological dimen- larged field sessions of the type that has tradi-
tionally characterized American archaeological
sions need little explanation because anthropolo-
gists are familiar with the problems of thedata na- collection. To be sure, the work may be
ture and stability of natural environment andneater,
in more attention may be given to stratig-
the physical and demographic human basis of
raphy, more classes of phenomena may be ob-
cultural systems. However, the social dimen- served and collected than in the field work of
sion is frequently excluded from considerations years past; yet the general methods of data col-
lection and observation remain unchanged. I
of adaptation. There is little need to belabor the
wish to argue that current lack of concern with
point that as the density and complexity of sep-
arate socio-cultural systems increase within thea development of planned research designs
major geographic zone, the cultural means generally
for obviates the recovery of data pertinent
articulating and "adjusting" one society with to questions which derive from current theo-
another become more complex. Certainly reticalthe interests. Investigatory tools must fit the
job; current field procedures were developed to
coping situations and hence the adaptive stresses
associated with a changing pattern of socio- provide data relevant to a limited number of
problems. Concern has been with problems of
political distribution within a major zone must
be considered when attempting to understand stylistic chronological placement and historical
changes in any given system (Gearing 1962). continuity between and among archaeologically
As long as "cultures" are defined in termsdefined
of units. The methodological tools devel-
oped for the investigation of such problems are
stylistic similarity, and the question of possible
differences in the material inventory of func- inappropriate for supplying information rele-
tional classes and in the internal structure of the
vant to our broadening research interests in
assemblage is unanswered, there is little possi-cultural processes.
bility of dealing realistically with questions of
process. It is a system that is the seat of process. METHODOLOGICAL PROBLEM AREAS
Because of these convictions I will frequently
Laymen frequently ask: "What are you dig-
mention the "regional approach" or the detailed
ging for?" I think most of us will agree that we
and systematic study of regions that can beare
ex-
digging to recover facts for the elucidation of
pected to have supported cultural systems. past
The cultures. In the absence of explicit state-
BINFORD ] ARCHAEOLOGICAL RESEARCH DESIGN 427
ments concerning the kinds of facts which notar-

excavate every location which yields surface
chaeologists hope to obtain, we can only assume
indications. What are the means whereby we
that we know how to recover the "pertinent" select certain sites for excavation and not others?
ones. Most of us will agree that this is not true.
A quick review of 37 regional reports spanning
the period from 1954 to 1963 failed to reveal a
I not only have been unable to use other investi-
gators' data, but I have also frequently foundsingle exposition by the authors as to the criteria
my own data lacking in many important "facts"they utilized in selecting sites for excavation. A
-facts which could have been collected had I typical statement may read as follows:
been aware of the questions to which the given During this time a total of 51 sites was located in sur-
observation was relevant. For instance, I re-
vey work and 13 of the more important sites were exca-
cently wanted to demonstrate that most ofvatedtheto some extent.
sites in a particular area were located adjacent
It is my impression that there is no single set
to streams. This was impossible because I ofhad criteria for selecting "important" sites. Some
no data as to where the archaeologist reporting
archaeologists select sites because they represent
on the area had concentrated his survey efforts.
a time period. Others are selected because they
Was the failure to report sites in areas not adja-
are large and productive. Certainly some have
cent to streams the result of sites being absent,
been excavated because they were accessible to
or was it simply a lack of investigation in those
modern roads. Despite the lack of systematic
areas not adjacent to streams? In another in-
statement, it is repeatedly mentioned that sites
stance, I wanted to compare the relative density
are representative or that they are large and
of Middle Woodland sites in two major river
yield much material. Less frequently, economy
basins in order to make statements about the
is cited as a reason for selecting a particular site.
relative occupational intensity in the two areas.
Although it is not commonly expressed, we may
This was impossible because I was unable to
generalize that archaeologists want representa,
determine whether or not the reported differen-
tive and reliable data within the bounds of their
tial densities were the result of differences in the
restricted time and monetary resources. This
intensity of survey, presence or absence of forest
is practically the definition of the aims of mod-
cover affecting the likelihood of sites being recog-
ern sampling procedures. Sampling, as used
nized, or differential aboriginal use of the valleyshere, does not mean the mere substitution of
under consideration. Such uncertainties make
a partial coverage for a total coverage. It is the
it obvious that we are concerned with answering
science of controlling and measuring the relia-
questions for which our research designs, field
bility of information through the theory of prob-
methods, and reporting procedures are not ade-
ability (Deming 1950: 2). Certainly we are all
quate to supply the "pertinent facts." Such
aware that we must substitute partial coverage
a situation cannot be prevented entirely, but we
for complete coverage in our investigation. Giv-
can strive to devise techniques for gathering the
en this situation, there is only one currently
facts which are pertinent to questions currently
known means for accomplishing coverage so
being asked of our data. As the general theo-that the results can be evaluated as to their re-
retical development within archaeology goes for-
liability in representation of the population in-
ward, more and more facts previously ignored
vestigated. This is through the application of
will be recognized as important and pertinent.
sampling theory in the development and execu-
We can look forward to continued concern with
tion of data-collecting programs.
keeping our investigatory tools sufficient to the
task; in short, we will be increasingly involved SAMPLING
in the development of new and improved re-
search designs and methodologies by means of In this discussion of sampling we shall intro-
which they may be operationalized. duce certain terms that are used by writers on
One clue to a "methodological problem area"sampling and attendant statistical problems. A
in our current practices can be found in anotheruniverse is the isolated field of study. In most
question which laymen are apt to ask: "How do cases archaeological field work is conducted
you know where to dig?" My answer to such awithin a universe of territory, a universe spa-
question is that we dig where there are surface tially defined. A population consists of an aggre-
indications of past occupations or cultural ac-gate of analytical units within the universe so
tivity - a seemingly accurate answer, yet we do that, at least in principle, each unit may be as-
signed a definite location for a given unit of (1) The population to be sampled and the units com-
time. The population has a distribution in space posing it must be clearly defined so that there will be no
question as to what the sample represents.
consisting of the aggregate of individual loca-
tions (Duncan and others 1961: 21). In addi- (2) A universe partitioned by a frame composed of
many small units is preferable to one composed of fewer
tion to a distribution, we can speak of the spa- but larger units. This is a safeguard against accidental
tial structure of a population. Structure sug- inclusion of an unrepresentative amount of "hetero-
gests a pattern of interrelationships among dis- geneity" in any given sample.
tinguishable parts of an organized whole (Dun- (3) The units of the frame should be approximately
can and others 1961: 2), and for our purposes equal in size. This eliminates bias which could result
the spatial structure of archaeological popula- from a systematic relationship between the structure and
the size of the population.
tions derives from the complex interrelation-
ships between people, activities, and material (4) All units should be independent of each other
so that if one is drawn for sampling, it will in no way
items within a cultural system. In addition, we affect the choice of another.
may speak of the form of a population, which is
(5) The same units should be used in sampling, tabu-
the nature and quantitatively variable consti- lation, and analysis. A sample of mounds is of no use if
tution of subclasses and the relative frequency generalizations about general site distributions are being
of analytical units. attempted.
The application of the method of probability (6) The universe must be present or cataloged so that
every unit in it is listed or can be given an identifying
sampling presupposes that a universe can be sub-
symbol to be used during the drawing of a sample. For
divided into distinct and identifiable units called instance, a grid system is established with 12 ten-foot
sample units. These units may be natural units, squares on a side, and 20 of the 144 squares are chosen
such as sites or individual projectile points, or for excavation by a random method. Later, it is decided
to extend the grid system six more squares in one direc-
they may be arbitrary units, such as 6-inch levels tion. The enlarged system does not have the random
in an excavation, or surface areas defined by a character of the sample drawn under the frame defined
grid system. Regardless of the basis for defini- by the original grid system.
tion, the application of the method of probability (7) The method of drawing the sample should be
completely independent of the characteristics to be ex-
sampling presupposes the availability of a list of amined.
all the potential sample units within the uni-
(8) In order for the sample to remain random, every
verse. This list is called the frame and provides unit drawn must be accessible. For instance, in the case
the basis for the actual selection of the sample where a site has been selected for sampling and the prop-
units to be investigated. The frame varies with erty owner refuses permission to dig, inaccessibility biases
the nature of the archaeological population the sample. This is particularly true because refusal of
permission may be related to ideas of the "value" of
under investigation. When a population of sites materials on his property relative to those on others'
is sampled, the frame is normally a list of sites property.
within a stated universe, such as the alluvial With these principles serving as a background,
bottoms of the Rock River between two specified we can turn to a discussion of types of sampling
points. When partial coverage of a population and their ranges of applicability to archaeologi-
within a stated universe is attempted, the sample cal investigation. Although there are many
units are selected from the frame so that all types of sampling, only two will be discussed:
units of the frame have an equal chance of be- simple random sampling and stratified sampling.
ing chosen for investigation; the selection is gov-
erned by the "laws of chance" alone, maximiz- SIMPLE RANDOM SAMPLING
ing the reliability of the sample. This is by far the simplest of the methods of
Before we approach the subject of different probability sampling. It implies that an equal
methods of probability sampling and their range probability of selection is assigned to each unit
of application in archaeological research, cer- of the frame at the time of sample selection.
tain principles which underlie and guide the re- The term random refers to the method of select-
search design aimed at the proper and efficient ing the sample units to be investigated rather
execution of sampling techniques will be men- than to the method of investigating any given
tioned. This presentation is adapted from a sec- unit. A practical procedure for selecting a ran-
tion entitled "Types of Sampling" by Parten dom sample is by utilizing a table of random
(1950). numbers (Arkin and Colton 1957: 142). The
procedure is to (1) determine the number of

economy is greatly affected by the number of
subclasses into which the recovered data will be
units in the frame, identifying them serially
from one to (n); (2) determine the desireddivided.
size For instance, in the case of a sample
of the sample, that is, actual number of units
of to
ceramics composed of 100 sherds, it is likely
be investigated; (3) select the required series
thatofif only two "types" are represented, the
numbers from the table of random numbers sample may be sufficient to give a fairly reliable
(the required series is the number determined estimate of the relative proportions of the two
sufficient to constitute a representative sample of
types in the population. On the other hand, if
the population); and (4) investigate those units within the sample there are 15 "types," then the
in the frame that correspond to the numbers reliability of the sample as an estimate of the
drawn from the table of random numbers. It relative proportions of the recognized types in
cannot be overemphasized that this is a tech-the parent population is very low. A further
nique for selecting the units to be investigated caution applies to samples drawn from multi-
and does not refer to the procedures used incomponent sites where there are clearly several
gathering the data. recognizable populations of separate historical
origin. In the latter case, some of the subclasses
STRATIFIED SAMPLING
are independent of one another, and any given
It can be shown that the precision of a sam-subclass may be representative of only one of the
ple depends upon two factors: (1) the size of multiple historical populations. Also, in the lat-
the sample and (2) the variability or hetero- ter case, sample size must be determined by the
geneity of the population being sampled. If we relative frequencies of the smallest independent
subclass. In other words, sample size should be
desire to increase the precision, aside from in-
creasing the sample size, we may devise meanslarge enough to give reliable measures of the
which will effectively reduce the heterogeneity smallest important breakdowns made within the
of the population. One such procedure is known sample. If the population is relatively homo-
as the method of stratified sampling. The pro-geneous, then sample size may be relatively
cedure is to partition the universe or divide it small and still reliable. On the other hand, if
into classes, each of which is treated as an inde- the population is heterogeneous, more observa-
pendent sampling universe from which simpletions are needed to yield reliable data.
random samples are drawn, following the meth- This discussion of sampling is very elemen-
ods outlined above. This procedure has a num-tary, but enough has been presented to serve as
ber of advantages. Classes may be establisheda basis for evaluation of the types of observa-
with regard to different variables that one wishestional populations which archaeologists investi-
to control, which makes possible the reliablegate and their investigatory peculiarities as a
evaluation of variability in other phenomenabasis for further discussion of research design.
with respect to the class-defining variables.
As has been suggested, the size of the sampleBASIC UNITS OF ARCHAEOLOGICAL OBSERVATION
is an important factor for consideration in striv-
Albert Spaulding has provided us with a
ing for reliability in the sample as representative
classic statement of what we as archaeologists
of the population. An optimum sample is one
are doing, thereby setting forth an operational
which is efficient, representative, reliable, and
definition of the field of archaeology. Spauld-
flexible. The sample size should be small enough
to avoid unnecessary expense and large enoughing's introductory statement is reproduced here
as a point of departure for further discussion
to avoid excessive error. To arrive at a sample
size which is considered optimum in terms of of archaeological data collection.
the above criteria, there are a number of factors A science deals with some class of objects or events in
which must be considered, each largely inte-terms of some specified dimensions of the objects or
events. The simplest (and most elegant) of the sciences,
grated with the other rather than independent.
mechanics, has all physical objects as its center of at-
Each will vary, in different sampling situations, tention, and the dimensions of these objects as studied are
as to their relative importance in influencing de-length, mass, and time; i.e., roughly speaking mechanics
cisions about the appropriate sample size. Of has three kinds of measuring instruments: a yardstick, a
particular importance to archaeologists is theset of scales, and a clock. The interrelationships and trans-
formations of measurements with these scales is the busi-
realization that the size of the sample necessary ness of mechanics. It is clear that prehistoric archaeology
to meet normal requirements of reliability and
also has a class of objects, artifacts, as its center of atten-
tion. The concept "artifact" presupposes the idea of cul-

say that such information is not crucial to inter-
ture, which I will treat as a given. Thus we can define an
pretation of the formal properties, only that the
artifact as any material result of cultural behavior. Since
cultural behavior is our ultimate referent, it followsformal
that properties themselves are independent of
the matrix and provenience associations.
we are interested in only those properties, characteristics,
The
aspects, and attributes of artifacts which are the result of form of the cultural item may vary in
or have a significant relation to cultural behavior.terms
What of the function of the item as an element
are the dimensions of artifacts whose interrelationships
in the cultural system, for example, technologi-
are the special business of archaeology? Plainly there are
two in the strict sense of dimension: time and space.callyWe (manufacturing techniques and raw ma-
want to know where and when artifacts were made, terials)
used, or stylistically.
and deposited. Plainly there is another class of dimen-
The sampling universe for the investigation of
sions fundamental to archaeological study; the many di-
populations
mensions which are sets of physio-chemical properties of
of cultural items is necessarily the
the artifacts. We can group them for convenient site.
refer-The sampling and field-observation proce-
dures utilized do not affect our ability to analyze
ence under the label, formal properties, and collectively
as the formal dimension. We are now in a positionitems
to de- formally, but they greatly affect our ability
fine archaeology as the study of the interrelationships and
to study the distribution, form, and structure of
transformations of artifacts with respect to the formal,
a population of cultural items. It will be re-
temporal, and spatial dimensions. As a footnote, formal
membered
and spatial attributes can be observed directly, but tem- that a population necessarily has spa-
poral attributes are always inferred from formal tial and
attributes both in its distribution and its
spatial attributes. Indeed strictly speaking artifacts are
structure. Sampling control is therefore neces-
objects and do not have temporal attributes - they mere-
sary to provide data for the description of pop-
ly exist. But artifacts do imply events, and events do have
the property of occurring at a definite time, so whenulations
we of cultural items. We as anthropolo-
gistswe
speak of the temporal attributes of an artifact, hope to be able to assess the range of formal
variability
really refer to an inference about some event or process in classes of cultural items and to
implied by the formal and/or spatial attributes of the arti-their
study distribution and population struc-
fact. This leaves us with describing and ordering formal
ture in terms of spatial clusters of quantitatively
and spatial attributes as the primary task. These are the
variant
empirical data of archaeology and this describing and class associations. It is necessary to de-
fine accurately the range of formal variation
ordering are prerequisite to the chronological inferences
within classes of cultural items as well as rela-
(Spaulding, mimeographed version, revised 1960: 437-9).
tive frequencies among recognized classes with-
Artifacts as a class of phenomena represent
in the population. While excavating, we have
a number of different types of populations which
are definable in terms of their spatial, formal,precise
no or knowledge of the boundaries of the
population of cultural items being investigated,
spatial and formal attributes taken in combina-
tion. Because of the different nature ofand we generally sample in terms of areal units
arti-
designed to cover the territory defined by the
factual populations, investigation of the several
presence of artifacts. We are sampling "artifact
recognizable populations must differ as regards
space" as a means to both definition and segre-
the appropriate means to provide the necessary
information for their formal, spatial, andgation
tem-of populations of cultural items. We can
only accomplish this through exercising tight
poral analysis. The several classes of artifactual
populations recognized here will be termed spatial
types controls for gaining information neces-
sary
of observational populations, since it is argued to the analytic determination of what cul-
that different sampling techniques, and tural
henceitems are, spatially and temporally clust-
ered one with another and with other arti-
research strategy, are necessary for the investiga-
tion of each. factual materials. Such insights are a clue to the
"role played" by various items in the operation
POPULATIONS OF CULTURAL ITEMS of extinct cultural systems. Similarly, the same
control is necessary to the definition of the spa-
Cultural items are discrete entities, the formal
tial structure and form of the population. We
characteristics of which are at least partially the to utilize techniques which will insure re-
want
result of cultural activity or events. A further
liable and representative data regarding the
range
qualification is that the formal characteristics of of formal variability within a given sub-
the item are not altered through removalclass from of cultural items, the content of the popu-
their matrix; they are transportable andlation may of cultural items defined by the relative
frequencies of the recognized classes, and the
be formally analyzed without recourse to infor-
mation about their provenience. This is not structure
to of the population defined by the spa-
tial structure of between-class associations of Sources of error which frequently arise in

items and other classes of artifacts. sampling such populations are (1) incomplete
Error arising in the sampling of populations
and nonrepresentative coverage of the "artifact
space" so that, while the number of recovered
of cultural items normally results from (1) in-
complete and nonrepresentative coverage of thefeatures may be large, there is no way of demon-
universe; (2) failure to partition the universe strating or determining whether the between-
so that a single undifferentiated collection class
is frequencies are representative of the popu-
made, thereby excluding the possibility of inves-lation present; (2) samples are far too small to
tigating the homogeneity or heterogeneity of the allow adequate evaluation of the variability rep-
population; or (3) samples are far too small resented in a given class of features or to yield
to allow the adequate evaluation of the varia- reliable estimates of the between-class relative
bility represented in a given class of cultural frequencies.
items or to yield a reliable estimate of the be-
tween-class relative frequencies. POPULATIONS OF CULTURAL ACTIVITY
Loci: SITES
POPULATIONS OF CULTURAL FEATURES
The site is a spatial cluster of cultural features
Cultural features are bounded and qualita- or items, or both. The formal characteristics
tively isolated units that exhibit a structural as- of a site are defined by its formal content and
sociation between two or more cultural items the spatial and associational structure of the
and types of nonrecoverable or composite mat- populations of cultural items and features pres-
rices. The cultural feature cannot be formally ent.
analyzed or at least formally observed after its
(1) Sites vary in their depositional context.
dissection in the field. Many of the formal ob-
servations must be made while the feature is be-
Sites exhibiting primary depositional context
have not been altered in their formal properties
ing excavated. Features include such classes of
except through the natural processes of the de-
remains as burials, mounds, structures, pits, and
cay of organic material, or the physico-chemical
hearths. Formal variations among cultural fea-
alteration of features and items since the period
tures are dependent upon (1) their functions
of occupancy. Sites exhibiting secondary deposi-
within the represented cultural system, (2) tech-
tional context are those whose formal character-
nology in terms of the raw materials utilized in
istics, defined in terms of soils, features, and
their production, (3) alterations occurring as a
items, have been spatially altered through physi-
result of their participation in other natural sys-
cal movement or deletion from the loci. Some
tems, for example, organic decay, (4) their cul-
or possibly all of the original associations be-
tural history (how often they were parts of suc-
tween the various classes of artifacts have been
cessive cultural events, resulting in their repair,
changed. This disruption in the structure of the
secondary modification, and destruction), and
site may have occurred through the agency of
(5) stylistic variation. Unlike cultural items,
erosion, geophysical changes, or through destruc-
the cultural feature cannot be formally defined
without precise and detailed observation andtion as a result of later cultural activity. Sites
with primary depositional context yield the most
"analysis" in the field. The field investigator
must at least make decisions as to what attri- complete archaeological record. However, sites
with secondary depositional context must fre-
butes are culturally relevant and meaningful
quently be studied in order to understand the
prior to beginning field observation and record-
regional distribution of activity loci.
ing. This adds an additional field burden to the
normal exercise of sampling control character- (2) Sites vary in their depositional history.
istic of the investigation of cultural items. The culturally dependent characteristics of a
As in the case of cultural items, the sampling site may have been the result of a single short-
universe for populations of cultural features isterm occupation, a single long-term occupation,
the site. Similarly, our sampling procedures multiple occupation over a rather limited tem-
should insure reliable and representative dataporal span, multiple occupations over an ex-
tended period of time, or combinations of all
regarding (1) the variability within any class of
of these.
features, (2) the formal content of the popula-
tion of features, and (3) the structure of the (3) Sites vary in their culture history. Sites
population of cultural features. exhibiting a complex depositional history may
or may not exhibit a complex cultural history. is the methodological aim of sampling a uni-
A site could be repeatedly occupied by repre- verse of sites regionally defined.
sentatives of the same stable socio-cultural sys- Two major sources of error arise in the inves-
tem for the same purposes, in which case there tigation of site populations. The first source of er-
may be a complex depositional history with a ror is incomplete and nonrepresentative cover-
simple cultural history. Similarly, a site with a age of the range of variation represented among
simple depositional history may exhibit a com- sites within the universe. This arises inevitably as
plex cultural history, for example, an extended a result of the "selection" of sites for investi-
long-term occupation spanning a period of major gation on the basis of criteria other than those of
structural changes in the cultural system. A the method of probability sampling. For instance,
single locus may be sequentially occupied by sites are frequently selected because of a high
social units of different socio-cultural or socio- density of cultural items almost to the exclusion
political units, adding to the complexity of the of sites with low density. The density of cultural
cultural history of the site. items at a site is a formal attribute of the specific
activity loci and is only relevant to the selection
(4) Sites and areas within sites vary function- of sites for excavation as an attribute in a pro-
ally. Since sites are the result of cultural activi-
visional site typology. A given universe may
ties performed by social units within restricted
have very few sites with dense concentrations
spatial bounds, we would expect them to vary
of cultural items, while the number of sites ex-
formally as a function of the activities of the
hibiting less-dense concentrations may be quite
social units represented. It is a known and de-
high. In this case, the sample of sites for investi-
monstrable fact that socio-cultural systems vary
gation must be composed of a proportionally
in the degree to which social segments perform
higher number of sites exhibiting low densities
specialized tasks, as well as in the cyclical pat- of cultural items. The second source of error is
tern of task performance at any given location.
failure to sample with sufficient intensity to
These differences have spatial correlates with
yield a reliable measure of the variability present
regard to the loci of task performance; hence we in the population. Inadequate sample size meas-
expect sites to vary formally and spatially with ured by the number of investigated sites is one
regard to the nature of the tasks performed at of the major sources of error. Ideally, a sam-
each, and the social composition of the units ple of sites should be adequate to represent the
performing the tasks. formal range of variability in site form, the rela-
All possible combinations of the above-men- tive frequencies of recognized site types, and
tioned basic forms of variation may occur at their spatial structuring within the universe.
sites which archaeologists investigate. Archae-
POPULATIONS OF ECOFACTS
ologists must be prepared to make the pertinent
observations needed to define the form and In addition to the investigation of cultural
structure of the populations of artifacts and cul- items, features, and activity loci, we must sam-
turally relevant nonartifactual material present, ple populations of ecofacts. Ecofact is the term
and to isolate the form and structure of his- applied to all culturally relevant nonartifactual
data. Cultural systems are adaptive systems,
torically different archaeological assemblages
and in order to understand their operation and
represented. Unlike populations of cultural
the processes of their modification, we must be
items or features where the normal universe
in a position to define their adaptive milieu. All
is the site, the sampling universe for populations those elements which represent or inform about
of sites is of necessity a region. Once the archae- the points of articulation between the cultural
ologist has determined the relative homogeneity system and other natural systems must be sam-
of the sampled site population as regards the pled. This is an extremely important phase of
historical and functional nature of the archae- archaeological data collection and is accom-
ological assemblage present, he is in a position panied by many field complications in terms of
to consider the nature of the site as a whole and methods of observation and sampling. The gen-
to classify it within a typology of sites (based on eral class of ecofacts can be broken down into
the attributes of both the form of the artifactual many subclasses representing different popula-
elements present and the structure of their spa- tions, such as pollen, soil, and animal bone,
tial and formal associations). Such an approach each with specific attendant sampling problems.
concerned themselves with vertical spatial

However, for the purpose of this presentation,
we will consider ecofacts as a single population
analysis, and the search has been for stratified
sites in which a limited "test pit" will yield a
which, in general, requires certain methodologi-
stylistic
cal considerations distinct from the problems as- sequence that may be used to develop
a regional chronology. This results in "cultures"
sociated with sampling artifact populations.
being isolated on a regional basis and defined
BASIC SAMPLING UNIVERSES
largely in terms of the stylistic characteristics of
Although we have recognized fourcultural
major items. The resulting information is in-
types of observational populations, eachsufficient
differ-for the structural definition of arti-
fact assemblages and site typologies in precise
ing in the way it must be observed and sampled,
terms. in
there are only two basic sampling universes Cultural "taxonomy" remains almost ex-
clusively
excavation or field work, the region and the site.in stylistic terms.
On
Populations of sites must be investigated withinthe other hand, there has been inordinate
a universe defined in spatial terms, theinterest
region. in certain classes of cultural features,
Populations of cultural items and features suchmust
as burial mounds and platform mounds.
be investigated within a universe defined It bywas
the early recognized that such mounds were
excellent
bounds of artifactual distribution at a given lo- "mines" for exotic and artistically
cation, the site. Ecofactual populations pleasing
may beobjects and were therefore attractive
sampled within both universes, depending to untrained
on investigators and relic hunters.
the types of information desired. If cultureWork
typesin mounds, whether prompted through
humanistic
are to be defined, it is essential that we isolate a interests or through the "salvage
reliable and representative sample of the motive,"
popu-has contributed inordinately to our
lation of sites characteristic of a given "sample"
culture.of artifactual data which serves as the
For adequate definition of types of activitytaxonomic
loci basis of many archaeologically de-
we need a reliable sample of populationsfined cultures. This lack of representative data
of cul-
tural items and features assignable to anyplus inadequate information on the form of the
given
occupation. In order to obtain such informa- features is a real and limiting bias in the data
tion, we must have well-planned research currently
de- available for study. Data have been
signs rooted in the application of probability gathered in terms of problems which concen-
sampling procedures. trate investigations on populations of cultural
items at the expense of and to the exclusion of
LIMITATIONS OF CURRENT PROCEDURES
cultural features. Investigation has also been
It is my impression that archaeologists concentrated
have on particular types of obvious fea-
not consciously aimed at sampling populations tures, such as mounds, and there has been very
of sites. They have concentrated on collecting little awareness that the aim of archaeological
"samples" of cultural items within regionally investigation is the definition of the structure of
defined universes. The sites have been treated an archaeological assemblage in addition to its
largely as "mines" for such items. In excep-content. These factors have contributed greatly
tional cases, where sites have been intensively to our current inability to deal systematically
with archaeological data.
investigated and populations of cultural features
studied, there is little attempt to analyze theCurrent interests demand that we do not per-
population of cultural items with regard to its petuate these limitations in our methodology.
spatial structure or form, while inordinate at- We must approach our work with the methodo-
tention is frequently given to describing the logical ideal of sampling a spatial universe, re-
"norms" of recognized formal subclasses within gardless of whether it is conducted under large-
the population. When cultural features are in- scale regional research programs or over an ex-
tended period of time through a series of small-
vestigated, they are usually reported cartographi-
cally, with little attempt to conduct a detailedscale investigations. Such sampling is aimed at
obtaining a reliable and representative sample
formal analysis aimed at the description of types
of features. One rarely finds a report in which of the range of variation in formal-structural
terms of sites within a given region. Selection
correlations between the spatial structure of pop-
ulations of cultural features and items have of sites for excavation should be made on the
been attempted as a matter of "standard" pro- basis of some method of probability sampling as
cedure. For the most part, archaeologists have the best means of insuring that the expenditure
of time and money in excavation will yield the these segments were articulated into a func-
desired information. Sites selected for excava-tional cultural system. We want to know the
tion must be investigated so that they can demographic
be basis and how it varies with respect
formally defined from the standpoint of the to isolated structural changes in the cultural sys-
tems. In short, we want to know all we can
nature of the populations of cultural items pres-
ent, but equal attention must also be given about
to the structure and functioning of the ex-
the population of cultural features. This is the
tinct cultural systems and how they relate one
only way to approach the necessary task of de- to another as regards processes of change and
veloping a site typology in functional and struc-evolution.
tural terms, an absolute necessity for the defini-The initial problem is the location of the
tion and isolation of the archaeological structure
various loci of past cultural activity within the
of extinct cultural systems. The latter is judgedregion. This phase of the work should be di-
a necessary step toward the scientific investiga-rected toward determining the density and dis-
tion of cultural processes. tribution of activity loci with respect to classes
of ecofactual phenomena, such as plant com-
A HYPOTHETICAL RESEARCH DESIGN
munities, physiographic features, and soil types.
In an initial attempt to think through some In order to accomplish this task, there is only one
appropriate procedure short of complete cover-
of the practical problems associated with the de-
sign and execution of a research program whichage, a procedure rooted in some form of probabil-
attempts to operationalize some of the sugges- ity sampling. One suggested approach is to strat-
tions advanced thus far, I will present a "hypo-ify the regional universe on the basis of ecofactu-
thetical" research program. Hypothetical aliscriteria judged desirable to control, such as soil
placed in quotes because many of the ideas and types. If we assume for purposes of presentation
problems discussed are the result of work cur-that soil types have been decided upon for the
areal stratification, in most cases the bounds of
rently being undertaken in the southern part of
the various soils will be defined fairly accurately
Illinois, specifically in the Carlyle Reservoir. Re-
gardless of the projected implementation on
ofa soil map, and we can simply determine the
extent of each in square miles, acres, or other
many of the ideas set forth, the program remains
hypothetical because the suggestions are un- appropriate units. Having accomplished this,
tried and undemonstrated. It is hoped that by we can impose a frame within each sampling
stratum (areas of common soil type). It will be
presenting these ideas in the form of a research
"model," others may gain a clearer understand- remembered that a universe partitioned by
ing of what is intended by the application of many small sample units is preferable to one
probability sampling approaches in field work. fewer but larger units, and that the units
with
It is further hoped that this model can serveof as the frame should be approximately equal in
a "whipping boy" for the improvement and fur- size. Using these guides, we can impose a grid
ther development of field methods and the exe- system over the areas of the various soil types.
cution of well-planned research designs. The actual size of a given unit in the frame
Let us assume that we are given the task of would be determined by considerations of sur-
investigating the prehistoric remains withinvey a logistics and the need to have multiple but
region. Our aim is to determine with the great-also practicable units for investigation. For pur-
poses of presentation, it is assumed that the grid
est degree of precision and reliability the nature
of the extinct cultural systems represented for is composed of squares equaling one-half square
the entire range of human occupation. We mustmile. We would then count and enumerate
face the problem of isolating the variable cul- each unit in the separate frames for each sam-
tural items, cultural features, and sites of ac- pling stratum (soil type). The next methodo-
tivity for the cultural systems represented. logical
In consideration is arriving at a "sample
addition, we must gather ecofactual data as a size." This can be quite complicated. For pur-
basis for understanding the way in which the poses of argument, it will be dismissed and we
extinct cultural systems participated in the re- will assume that a 20% areal coverage within
gional ecosystems of the past. We want to know each sampling stratum has been judged sufficient.
the internal structure of the systems, the degree The next step is to draw the sample for each
of structural differentiation and functional spe- sampling stratum, and this may be accomplished
cialization of the social segments, as well as howby use of a table of random numbers. The sam-
pling units within each frame will then populations

be com- of cultural items and features pres-
ent.
pletely surveyed for purposes of locating Limited data are obtainable through sam-
sites.
What are the advantages of such a proce- pling populations of cultural items present on
dure? If executed under ideal conditions, the itsurface
will of a site as well as through exercis-
permit the objective evaluation of siteing density
spatial control over the "context" of the arti-
in terms of ecofactual controls and also factual provide populations as regards ecofactual data.
data relevant to summary statements This about the
information provides the classes of attri-
intensity of past activities in the region as a in classifying sites. There are three
butes utilized
whole within definable limits of error. The mainpro- attribute classes which can be normally
cedure will also permit the concentration controlled
of ef- through the use of surface-sample
forts on intensive study areas, making the data: the size and density of the cluster of cul-
logisti-
cal expenditure less than if the entire region
tural items, the formal constitution of the popu-
were surveyed in a haphazard fashion.lation In addi-
of cultural items, and the degree of sty-
tion to these advantages, it eliminateslistic "hidden
and functional homogeneity of the popu-
bias" in the form of differential attention lation. paid
to ecological situations which the investigator The methods utilized to control the attributes
"feels" were preferred by prehistoric inhabitants. of size and density will also allow us to partition
By following such a plan it is possiblethe to population
dem- of cultural items and speak of
onstrate the ecological preferences of past the occu-
relative densities and of its formal classes,
pants of the region. that is, the spatial structure of the population.
On the other hand, there are certain prob- class of data, largely ecofactual, can
A further
lems which arise with any attempt to sample and must in be controlled. This is the topographic
this manner. Of primary importanceand are the
physiographic nature of the location. Such
conditions of the area itself in terms ofinformation the type can be obtained at the same time
of cover, presence or absence of modern that com-
the spatial controls are established for sam-
munities, and distribution of agricultural pling land.
the population of cultural items.
Such factors could variously affect one'sIn access order to control the relevant variables and
to the land for site locational survey as well
obtain theasnecessary data, we must have a num-
the relative efficiency of observation. These ber ofcom- sample units distributed over the area
plications are not new nor are they stumbling of the site and its immediate environs. These
blocks to the suggested procedures. They sampleare units must be rather evenly distributed
present no matter what type of research design
within and beyond the suspected bounds of the
we attempt to execute. The advantagesite. of Inthisaddition to establishing areal limits of
particular approach is that it provides a the method-
frame, we must determine the appropriate
ological frame of reference for documenting size of andeach sampling unit to insure the recovery
evaluating such bias. There are many of ways to
an adequate sample. A normal topographic
correct complicated sampling conditions. surveyWhen of the site will provide a basis for the
approaching the problem of locating sites as out-
notation of ecofactual data as well as for the
lined, it becomes imperative that the investi-
spatial control of the sampling frame and the
gator concern himself with the control of bias
location of sampling units.
resulting from differential survey conditions,
Some method of "systematic sampling" is sug-
something not generally considered under nor-
gested as being the most appropriate to surface
mal haphazard survey procedures.
sampling. Only the first unit is selected at ran-
Assuming that we have executed a research
dom and then others are selected in terms of a
plan as outlined, the next step is to define spa-
tially and sample initially the populations of pre-established interval (Vescelius 1960: 463).
cultural items present at each of the identified Systematic sampling ensures an equal dispersion
loci of cultural activity. This is prerequisite to of sample units, a desirable condition when den-
the evaluation of the formal characteristics of sities and aggregational analysis are attempted.
the sites themselves; the ultimate aim is a classi- There is a further advantage in that spatial con-
fication of activity loci as to their degrees of simi- trol on the placement of sampling units is easier
larity and difference. A working taxonomy of to maintain with an equal spatial unit between
sites is a necessary prerequisite to the selection them; thus, it is easier to lay out and identify
of sites for excavation and investigation of the the selected sampling units in the field.
A typical example of the execution of such cerning a the nature of variation that is observ-
program is given below: able in populations of surface-collected cultural
(1) Impose a frame over the area of the site in theitems and with techniques of probability sam-
pling. Are sites that exhibit similar size, density,
form of a grid system composed of sampling units of ap-
propriate size. and composition of cultural item populations
(2) Enumerate the sampling units in the frame from similar with respect to populations of cultural
one to n. Determine the necessary sample size and then features, depositional and cultural history, and
determine the appropriate sampling interval.
general function within the cultural systems rep-
(3) Consult a table of random numbers and draw the resented? If such a hypothesis were to be con-
initial sample unit. Then draw each sample unit separate
from the initial one by the designated sampling interval.
firmed, we would be in a position to generalize
far beyond the data derived from direct excava-
(4) Locate on the site the selected sampling units and
collect all cultural items within the bounds of that unit. tion and could make statements about settle-
ment systems based largely on surface-collected
This procedure can be speeded considerably data. This is not possible when sites are not
in open or cultivated areas by use of a "dog- treated within a sampling universe or when sur-
leash" technique. Each person who collects face data are not used in the generation of struc-
items has attached to his belt a cord of prede- tural hypotheses.
termined length to which is attached a stake. How do we actually go about the selection of
The stake is placed in the ground at the appro- sites for excavation? The following procedure is
priate location, and the person collects all the suggested:
cultural items within the radius of the circle de-
(1) Develop a taxonomy of sites based on formal at-
fined by the "dog leash." The location of sam- tributes investigated during the surface survey.
pling units can be determined quickly by means (2) Determine the relative frequencies and distribu-
of a tape and compass or with a transit. Such a tions of site types according to the original sampling
method is considerably faster than setting up a strata, for example, soil types.
grid and collecting items from a square unit, all (3) Stratify the population of sites into sampling
four corners of which must be defined. strata based on the typology further stratified in terms
of the original areal strata, that is, soil types.
Regardless of the particular procedure fol-
(4) Determine in terms of the time and funds avail-
lowed, application of the principles of proba- able what proportion of the total number of each site type
bility sampling to the collection of cultural items can be excavated to yield reliable information on their
from the surface makes possible the objective internal composition.
definition of the site in terms of density dines. (5) Enumerate each site in each sampling stratum
from one to n.
This permits objective comparison of sites in
terms of site size and item density, in addition (6) Consult a table of random numbers and draw the
appropriate sampling units designated by the random
to the form, homogeneity, and structure of the numbers.
population of cultural items present. On the (7) Proceed to excavate all those sites whose unit-
basis of such comparisons, we can arrive at a designator number was drawn from the table of random
provisional typology of the range of variability numbers.
in the population of sites within the regional The use of such a procedure can be justified
universe. Working hypotheses can be generated in a number of ways. First, it will be remem-
to account for the observable differences and
bered that the initial taxonomy based on sur-
similarities in form, density, and spatial struc- face-collected materials grouped sites judged to
ture, and these hypotheses can be tested by ex- be similar or different. Within each taxonomic
cavation.
class, sites are excavated to test the reliability of
A comparative study of information collected
this judgment and to further explicate the nature
through the application of sampling techniquesof the variability through more detailed investi-
provides a basic set of data for the construction
gation of populations of cultural items and fea-
of a stratified sampling frame of provisionaltures.
site Only by the use of such a procedure can
types within which selection of sites for excava-
we explicate the meaning of observed differences
tion can be made. This brings us to one ofinthe
the surface-collected material and thereby
major questions considered: how do we knowprovide the necessary information for confirm-
where to dig? I think that the answer toingthis
the validity of generalizations based on such
data. Secondly, the procedure insures an ade-
question logically rests with a methodology
quate and representative sample of the popula-
which attempts to test working hypotheses con-
tion of cultural activity loci within the defined generally inadequate as a sampling frame for
universe. It is a complete and unbiased across- cultural features. Ideally, we should have an
the-board investigation of the full range of X-ray for- machine which would allow us to locate
mal variability within the population. and formally evaluate the range of variation
The next phase of research planning ismanifest to in cultural features. Given such infor-
many the most important, and it is the phase mation, we could construct a frame and exca-
that has received most attention under the ru- vate features within each recognized formal class
bric of "field methods." Initially, it must beinrec- proportion to their relative frequency. Such
ognized that in excavation we are not sampling a procedure would be analogous to excavating
activity loci; we are sampling populations of selected on the basis of a previously de-
sites
cultural items, cultural features, and ecofacts fined frame of site types. Unfortunately, no such
at an activity locus that may or may not have X-raya machine exists, and we must attempt to
complex cultural and depositional history. We the desired sample by opening up areas
obtain
want data which will allow us to understand of the site in such a way as to (1) allow the
the historical aspects of the various occupations,recognition of the presence of cultural features,
as well as the functions of the occupations in the (2) provide a representative spatial coverage of
total cultural system represented. the universe in order to define the spatial distri-
If we view excavation as having a particular bution and structure of the features, and
(3) provide the necessary contextual data for
role in the scheme of data collection, it is reason-
able to think in terms of an excavational strate- the formal analysis of the recognized features.
gy. First, it must be kept in mind that all exca- This discourse is not intended as a discussion
vation is exploratory in addition to being a meth-of excavational techniques. However, a limited
od for securing samples. Some phases of exca- discussion of some widely utilized approaches as
vation may be parametric in the sense of it be- they relate to the general problems of research
ing possible to enumerate a sampling frame prior strategy seem to be in order.
to data collection, while other phases are ex- Ideally, once a site is selected as a unit in a
ploratory and sampling must be only provision-sampling frame of sites, it should be completely
ally parametric. Spatial control on a horizontal excavated. In such a case, there will be no ques-
axis makes possible the parametric definition oftion about one's ability to give parametric defini-
a sampling frame in terms of spatial units. Thistion to a sampling frame for cultural items com-
type of frame is ideal for investigation of the mensurate with the efficient investigation of cul-
homogeneity or heterogeneity of the popula- tural features. Complete excavation will insure
tion of cultural items, and it is appropriate for complete recovery of the entire record of past
exploratory work that seeks the solution of prob- activities at the given location.
lems of depositional and cultural history. How-In most cases it is impossible to undertake
ever, it is with excavation that we hope to ac- complete excavation of a site; only rarely are
complish the maximum correlational control funds and personnel available for such an un-
and thereby obtain the data that will allow re- dertaking, particularly when sites are large. One
liable interpretation of the internal variation inmethod frequently resorted to when faced with
the spatial structure of functional and stylistic a large site, or when the investigator is interested
classes of cultural items. Excavation will furtherin obtaining information concerning the consti-
provide well-documented and correlated sam-tution of the population of cultural features, is
ples of ecofacts, the basic data relevant to the to open up large "block" areas such as was done
nature of the local environment, and the way in at Kincaid (Cole 1951). A relatively large num-
which the represented social units were adaptedber of contiguous excavation units were opened,
to it. It is by correlating the distributions of cul-
and this resulted in the complete excavation of
tural items with different functional classes aof large "block" of the site area. This method in-
cultural features that insight into the "causes" sures recovery of the formal range of cultural
of differential distributions is obtained, and features present in any given block but, as nor-
hence understanding of the range, location, andmally implemented, does not insure that the ex-
nature of the various activities conducted at thecavated block is representative of the range of
site. Sampling frames designed solely to ob-features and activity loci present at the site. As
tain cultural items and provide information con- normally practiced, the block or blocks selected
cerning depositional and cultural history are for excavation are in the "core" area of the site
and therefore bias the sample toward features

tain specific questions with the most economical
and activity areas that were centrally located.
and expedient means.
As previously noted, our aim should be for ade-
As an example, initial discussion will center
quate, reliable, and representative data. Block
around a relatively small, single-component site
excavations as normally utilized do not supply
that lacks primary archaeological context below
this type of information. the plow zone (except, of course, cultural fea-
Test pits and test trenches are appropriate
tures). Initially, we want to know whether or
units of excavation when one is investigating
notcer-
there are clusters of differential density in
tain limited, formal properties of the site, but items, a clue to the possible location of
cultural
cultural
they are inappropriate to investigation of the site features. In addition, we want a com-
plete
as a whole. Test pits are by definition small, non-and unbiased sample of the population of
contiguous units. Such units are useful in pre- items in order to make judgments as to
cultural
liminary investigations of depositional problems
the "meaning" of demonstrable differential dis-
and as a means of solving site cultural history
tributions of recognized stylistic and functional
classes. Such a sample could be obtained by the
problems. They can also be profitably employed
in the collection of a dispersed number ofexcavation
sam- of a series of "test pits," the size and
ples of cultural items, but they do not normally
density of which would be determined by the
expose areas large enough to define and sample
estimated density of cultural items present. The
populations of cultural features. distribution of the pits would be determined by
Test trenches are excellent means of investi- some technique of probability sampling normal-
gating and defining problems of cultural andly executed within a grid frame. The plow zone
would be excavated and sifted for each of the
depositional history, but they have most of the
selected excavation or sample units. This meth-
limitations of test pits when sampling cultural
features. However, they frequently provide odology could be further implemented by com-
more information on the differential distribu- bining controlled surface collection from selected
sample units. The data collected in this way
tion of feature types if they happen to be opened
in sufficient density to "cross cut" major areas
should yield the desired information concerning
of the site. Data gathered from test trenches population form, structure, and content. Once
also have the advantage of being particularly this is accomplished the next phase of excava-
useful for analysis of item densities on a linear tion should be planned to yield the sufficient
axis. The limitations of test trenches are those controlled data on the population of cultural
of any technique which does not open up a large features present. In this particular case, where
contiguous area, and does not cover, in a repre-there is no primary archaeological context be-
sentative manner, the entire site. low the plow zone, we can most efficiently ac-
As test pits and trenches are normally utilized,complish our task by complete removal of the
they do not provide adequate data regarding the
plow zone with the aid of power equipment.
The result would be the exposure of cultural
population of cultural items because they are
features which could be mapped and excavated,
not normally distributed at random in sufficient
utilizing techniques designed to yield maximum
numbers over the site. A greater limitation is the correlational control.
failure to expose large contiguous areas, a neces-
The suggested excavation program would
sary condition for adequate sampling of cultural amount to a two-phase sequence. The first
features.
phase is designed to yield information about
Phase excavations seems to be the most appro- the population of cultural items present, where-
priate term to apply to the procedures which as the second phase would yield the desired in-
will be suggested as a means of overcoming some formation concerning the population of cultural
of the difficulties inherent in sampling the differ-features.
ent types of observational populations at a site. In many field situations the sites are more
The term implies that the excavation of a site complex, having multiple occupations with pri-
may involve several different excavational steps, mary archaeological deposits below the dis-
each largely dependent upon the results of the turbed plow zone. In such a case, a three-phase
earlier "phase" for the details necessary to the excavation program may be more appropriate.
proper planning and execution of the succeed- The initial phase would consist of opening up a
ing phase. Each phase is designed to answer cer-series of test pits selected for excavation on the
culture
basis of a random or systematic sampling historical aspects, more and more atten-
pattern
within a grid frame. This procedure,tion must
as in thebe given to maintaining maximum
earlier case, should provide the data necessary
correlational control. The intensity of sampling
for the reliable definition and isolation of
ordiffer-
the sample size needed to ensure adequate
ent stylistic and functional areas within data
theincreases
total with the heterogeneity of the uni-
site area. Next is the problem of sampling verse under investigation. This means that data
popu-
lations of cultural features. Since the site has necessary for justifying the step from one phase
several components, tight correlational control to another become expanded, and in general the
must be exercised to ensure the possibility nature
of of the appropriate phases changes with
correlating feature forms with forms of culturalthe complexity and form of the universe. The
more complex the site, the more complex the
items representative of discrete occupational epi-
sodes. An appropriate procedure is to employ excavational procedures, and the larger must be
block excavations as the second phase. Since thethe recovered samples for any given phase of
multicomponent nature of the site presumably excavation.
would have been recognized during the early This recognition provides the justification for
stages of the test-pitting phase, such a procedure
what I call the planning of an excavational se-
also ensures that the initial exploratory expo- quence. In areas where a number of sites have
sures will yield information obtained under con-been selected for excavation, the temporal se-
ditions of maximum correlational control. This quence of excavation can be very important in
hopefully makes possible the correlation of fea- promoting the efficient use of resources in both
ture types with old soil surfaces, which in turn labor and funds. The initial sites excavated
can be isolated and investigated in terms of should be the least complex, so that the chances
cultural-item content. With this type of infor- of making false correlations are diminished and
mation we should, assuming that the block ex- the maximum conditions obtain for observing
posures have been successful, be able to develop
the formal spatial structure of features and cul-
a formal taxonomy of features which can betural items. Once an understanding is gained
correlated with the variable populations of cul-
of the formal and structural characteristics
tural items representative of the separate occu-
which may be encountered, one is in a much
pations. Once such correlations are established,
better position to investigate a complex site
the third phase of excavation can begin, thewhere the nature of the variability may not be
"stripping phase." This is removal, by meansso clearly depicted. Informed excavation of a
of power equipment or by hand, without at-complex site can often greatly expedite its effi-
tempting to recover cultural items, of the com- cient investigation.
plete cultural deposit down to the level where It is hoped that by following this "hypotheti-
cultural features can be observed intruding into cal" research program, the reader has gained a
the natural. Once this is accomplished, theseclearer understanding of what is intended by
features can be mapped and excavated, usingthe argument that methods of probability sam-
techniques that will ensure maximum correla-pling are applicable on all levels of field investi-
tional control. The distributional data thus ob-
gation. By pointing out some of the complica-
tained will supplement that already collected tions, I trust an appreciation can be gained as to
and make possible the definition of activity areas
the potential which the application of proba-
and the general internal community structure bility sampling methods holds for improving our
representative of the separate occupations. data-collection methods. Such methods further
If we have been careful in planning and suc-provide a basis for a greatly expanded analysis
cessful in the execution of the three phasesof ofarchaeological data directed toward the defi-
excavation, we should have the data necessary nition of archaeological assemblages in struc-
for the demonstration of differences and simi- tural terms, ultimately with a view toward the
larities between occupations in terms of the for-
isolation and definition of extinct cultural sys-
mal, spatial, and structural composition of tems.
the
separate populations defined by both culturalOf equal importance is the recognition that
items and features. field work must not be conducted separately
Enough has been said to suggest what is in-from analysis. Running analysis is a necessary
tended by phase excavations. As the complex-part of feature description, and of even greater
ity of a site is compounded in depositional andimportance is the recognition that the results of
440 AMERICAN ANTIQUITY [ VOL. 29, NO. 4, 1964
running analysis largely serve as the basis for the cerning the internal structure and ecological
planning and decision-making regarding succes- setting of successive cultural systems. It is ob-
sive methodological steps taken in the execution served that under current programs of salvage
of a field program. Much ink has been spilt on archaeology and greater foundation support
the argument that the archaeologist as such is for archaeological research, archaeologists are
a technician (Taylor 1948: 43). Only in a very actually being given the opportunity to study
restricted sense can such a position be defended such regions. It is argued that, in spite of such
because the field archaeologist is forever making opportunities, our current practices largely ob-
decisions as to what are pertinent and relevant viate the recovery of data necessary to the study
"facts." Such decisions can only be made with of cultural process. The development of tech-
knowledge and understanding of the questions niques for the recovery of data in structural
being asked of the data. The field archaeolo- terms is believed to be crucial, for it is the struc-
gist must also be an anthropologist to make such ture of archaeological remains that informs
decisions efficiently and effectively. As Brew about the cultural system, and it is the cultural
(1946: 65) has argued that there is no single or system which is the seat of process.
even adequate taxonomy sufficient for "bringing Probability sampling is suggested as a major
out all the evidence," so I also argue that there methodological improvement which, if executed
is no sufficient set of field techniques. Field work on all levels of data collection in full recognition
must be conducted in terms of a running analy- of the inherent differences in the nature of ob-
sis and against a backdrop of the widest possible servational populations which archaeologists in-
set of questions to which the data are potentially vestigate, can result in the production of ade-
relevant. This is no technician's job. This is the quate and representative data useful in the study
job of an anthropologist specialized in the collec- of cultural process.
tion and analysis of data concerning extinct cul- Observational populations of cultural items,
tural systems. Only after the myth of simplicity features, and activity loci are recognized as hav-
which surrounds the training of field archaeolo- ing certain characteristics which demand differ-
gists is dispelled, and after more attention is ent treatment in both field observation and sam-
given to recovering information concerning the pling methodology. On the other hand, only
operation of extinct cultural systems as opposed two major sampling universes, regions and sites,
to the recovery of things, will archaeologists are recognized as appropriate to field investiga-
make significant advances in studies of cultural tions. Many of the limitations of currently
process. available data are believed to derive from the
failure to sample populations of activity loci
SUMMARY AND CONCLUSIONS
within a regional universe. Emphasis has been
It has been argued that as archaeologists on we
sampling populations of cultural items within
are faced with the methodological task ofaiso- regional rather than a site universe. This pro-
cedure has made impossible the structural defi-
lating extinct socio-cultural systems as the most
nition of populations of cultural items or the
appropriate unit for the study of the evolution-
study of activity loci from a structural point of
ary processes which result in cultural similarities
and differences. If we view culture as man's view. Consequently, our current understanding
extrasomatic means of adaptation, we must iso- of the prehistoric past is largely in terms of style
distributions and cultures defined in terms of
late and define the ecological setting of any given
socio-cultural system, not only with respect discrete
to traits and stylistic characteristics; this is
the points of articulation with the physical and certainly not a situation conducive to studies of
biological environment, but also with points cultural
of process.
articulation with the socio-cultural environ- The argument for planned and well-paced
execution of research design has been presented
ment. It is suggested that changes in the ecologi-
cal setting of any given system are the prime in the form of a "hypothetical" research pro-
causative situations activating processes of cul-
gram, along with a limited discussion of the tech-
tural change. niques and levels of applicability of probability
It is argued that the methodology most ap- sampling procedures. Problems attendant upon
propriate to the study of cultural process is a the recovery of structural information within
regional approach in which we attempt to gain both the regional and site universe have been
reliable and representative information con- made explicit in a number of examples of types
of sampling problem and excavational situation.

DEMING, WILLIAM EDWARDS
It is concluded that the design and execution 1950of
Some Theory of Sampling. John Wiley and
Sons, Inc., New York.
a research program is the job of an anthropolo-
gist, and only in a limited way can the field ar-O. T., RAY P. CUZZORT, AND BEVERLY DUNCAN
DUNCAN,
chaeologist be considered a "technician." Field1961 Statistical Geography. The Free Press of Glen-
work is an on-going process demanding method-coe, Glencoe.
ological decisions based on a running analysis of FRED
GEARING,
the data recovered from prior field work. 1962 It is
Priests and Warriors. Memoirs of the Ameri-
concluded that if we are to be successful in the can Anthropological Association, No. 93. Me-
collection of data relevant to studies of cultural nasha.
process, field work must be conducted within HAAG, WILLIAM G.

the framework of a well-planned research de- 1959 The Status of Evolutionary Theory in Ameri-
sign which provides for the application of proba- can History. In Evolution and Anthropology:
bility sampling techniques at all levels of investi- A Centennial Appraisal, edited by Betty J. Meg-
gers, pp. 90-105. The Anthropological Society of
gation. The field strategy executed within the Washington, Washington.
framework of the research design must be di-
rected by a well-trained anthropologist capablePARTEN, MILDRED
of making interpretations and decisions in terms 1950 Surveys, Polls and Samples: Practical Proce-
dures. Harper and Brothers, New York.
of the widest possible factual and theoretical
knowledge of general anthropology, and the SPAULDING, ALBERT C.
types of questions must be drawn up which his 1960 The Dimensions of Archaeology. In Essays in
the Science of Culture: In Honor of Leslie A.
data may be useful in solving. It is believed that White, edited by Gertrude E. Dole and Robert
modification of current practices along these L. Carneiro, pp. 437-56. Thomas Y. Crowell
lines is a necessary prerequisite for moving ar- Company, New York.
chaeology on to the level of development whichSTEWARD, JULIAN H.
Willey and Phillips (1958: 4-5) have called the 1960 Evolutionary Principles and Social Types. In
"explanatory level." The Evolution of Man: Mind, Culture and So-
ciety, edited by Sol Tax, pp. 169-86. The Uni-
ADAMS, ROBERT M. versity of Chicago Press, Chicago.
1960 The Evolutionary Process in Early Civilizations.TAYLOR, WALTER W.

In The Evolution of Man: Mind, Culture, and
Society, edited by Sol Tax, pp. 153-68. The Uni- 1948 A Study of Archaeology. Memoirs of the
versity of Chicago Press, Chicago. American Anthropological Association, No. 69.
Menasha.
ARKIN, H. AND RAYMOND COLTON
VESCELIUS, G. S.
1957 Tables for Statisticians. Barnes and Noble Inc.,
New York. 1960 Archaeological Sampling: A Problem of Statis-
tical Inference. In Essays in the Science of Cul-
BINFORD, LEWIS R. ture: In Honor of Leslie A. White, edited by Ger.
1962 Archaeology as Anthropology. American An- trude E. Dole and Robert L. Carneiro, pp. 457-
tiquity, Vol. 28, No. 2, pp. 217-25. Salt Lake 70. Thomas Y. Crowell Company, New York.
City.
WAGNER, PHILIP
BRAIDWOOD, ROBERT J.
1960 The Human Use of the Earth. The Free Press
1959 Archaeology and the Evolutionary Theory. In of Glencoe, Glencoe.
Evolution and Anthropology: A Centennial Ap.
praisal, edited by Betty J. Meggers, pp. 76-89. WHITE, LESLIE A.
The Anthropological Society of Washington,
Washington. 1959 The Evolution of Culture. McGraw-Hill Book
Company, New York.
BREW, JOHN OTIS
WILLEY, GORDON R. AND PHILIP PHILLIPS
1946 Archaeology of Alkali Ridge, Southeastern
Utah. Papers of the Peabody Museum of Ameri. 1958 Method and Theory in American Archaeology.
can Archaeology and Ethnology, Harvard Uni- The University of Chicago Press, Chicago.
versity, Vol. 21. Cambridge.
COLE, FAY-COOPER UNIVERSITY OF CHICAGO
1951 Kincaid, A Prehistoric Illinois Metropolis. Uni- Chicago, Illinois
versity of Chicago Press, Chicago. May, 1963

1964 Binford L. A Consideration of Archaeological Research Design

Uploaded by

Copyright:

Available Formats

You might also like

1964 Binford L. A Consideration of Archaeological Research Design

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

1964 Binford L. A Consideration of Archaeological Research Design

Uploaded by

Copyright:

Available Formats

A Consideration of Archaeological Research Design

Author(s): Lewis R. Binford

A CONSIDERATION OF ARCHAEOLOGICAL RESEARCH DESIGN*

ABSTRACT 4). A cultural system is a set of constant or

isolation and definition of extinct cultural extent

ments concerning the kinds of facts which notar-

procedure is to (1) determine the number of

tion. The concept "artifact" presupposes the idea of cul-

tial structure of between-class associations of Sources of error which frequently arise in

concerned themselves with vertical spatial

pling units within each frame will then populations

and therefore bias the sample toward features

of sampling problem and excavational situation.

process, field work must be conducted within HAAG, WILLIAM G.

1960 The Evolutionary Process in Early Civilizations.TAYLOR, WALTER W.

You might also like